sglang/ GB200 perf

dsr1-fp8-1k1k-max-tpt

concurrency 4,096 · 6 days ago

cron
passed
re-run
Commit

fix (#28597)

liusy58·6 days ago
PR #28597 · fix
Run
GitHub Actions27734450270-1
Slurm job5320
GPUs48·prefill 16 / decode 32
ISL / OSL1024 / 1024

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

40,960

duration

351

max_concurrency

4,096

mean_e2el_ms

33,646ms

mean_itl_ms

1,630ms

mean_tpot_ms

33.38ms

mean_ttft_ms

2,920ms

median_e2el_ms

32,680ms

median_itl_ms

1,656ms

median_tpot_ms

33.79ms

median_ttft_ms

1,270ms

num_prompts

40,960

output_throughput

107,616tok/s

p99_e2el_ms

54,475ms

p99_itl_ms

2,699ms

p99_tpot_ms

35.28ms

p99_ttft_ms

23,198ms

peak_output_tokens_per_s

130,223s

request_throughput

117tok/s

std_e2el_ms

5,243ms

std_itl_ms

382ms

std_tpot_ms

1.64ms

std_ttft_ms

4,823ms

total_input_tokens

37,769,666

total_output_tokens

37,742,239

total_token_throughput

215,310tok/s